HiMAP: Learning Heuristics-Informed Policies for Large-Scale Multi-Agent Pathfinding

Key Files

• test.py: Run pathfinding to see the Success Rate.
• config.py: Parameters for test data generation, etc.
• datagen.py: Test data generator.
• ./data/DHC_Data/: These test data is taken from DHC repository and is used for tests on DHC data. Each .pth file contains 200 different cases given the map specification and the number of agents.
• ./data/LargeScale_Data/: Self generated test data used for tests on large-scale data. Each .pth file contains 50 different cases given the map specification and the number of agents.

Citation

If you find our code or work helpful, please consider citing us:

@inproceedings{tang2024himap,
  title={Hi{MAP}: Learning Heuristics-Informed Policies for Large-Scale Multi-Agent Pathfinding},
  author={Tang, Huijie and Berto, Federico and Ma, Zihan and Hua, Chuanbo and Ahn, Kyuree and Park, Jinkyoo},
  booktitle={AAMAS},
  year={2024}
}